Enterprise AI is entering a new phase. For the past two years, organizations have focused on identifying use cases, running pilots and proving that generative AI can deliver tangible productivity gains. Increasingly, however, the conversation is shifting from deployment to operations.
How do organizations forecast AI usage before projects go live? How do they monitor consumption once employees begin using AI at scale? Who owns the budget? And how do businesses prove that AI is delivering measurable value rather than simply creating another operational cost?
The challenge is becoming increasingly significant. According to the recent Flexera 2026 State of ITAM Report, only 31% of organizations have accurate visibility into AI software, while 59% say wasted AI spend has increased over the past year. The report suggests AI investment is outpacing organizations’ ability to govern and optimize it.
Those questions are becoming increasingly relevant for ServiceNow customers following the company’s overhaul of its pricing model earlier this year.
Under the new ServiceNow pricing model, the company replaced its five-tier licensing structure with three new AI-native tiers – Foundation, Advanced and Prime – while embedding AI capabilities such as Now Assist more broadly across the platform.
Rather than treating AI as an optional add-on, the new model introduces AI consumption through “assists”, with customers purchasing annual usage pools that are consumed as AI capabilities are used. The change reflects ServiceNow’s broader strategy of making AI a core part of the platform, while also introducing new considerations around forecasting usage, governance and long-term cost management.
“We decided a while ago to, instead of having the top end product being the only ones who get AI, to democratize AI effectively. We put a sprinkling in every one of our products,” Adam Spearing, enterprise advisory lead at ServiceNow UKI, told NowBen.
Rather than reserving AI for customers purchasing its highest-value products, ServiceNow has embedded baseline AI capabilities across its licensing tiers, with additional functionality and consumption scaling as organizations increase their use of AI. The result is that customer conversations are beginning to evolve.
“We are seeing this pattern of people getting very excited… ‘I’m ready to go. Oh my God, how much is it? Am I going to use all this?’” said Spearing.
The Forecasting Challenge
Cloud computing forced IT departments to develop new disciplines around usage monitoring and cost optimization. AI is beginning to present similar challenges.
“As we look at the roadmap for the Control Tower… I want us to get to a point where we can give predicted usage on an AI project. I can turn around to you and say, if you implement that, it’s going to burn that number of tokens,” said Spearing.
Today, organizations can measure AI usage after deployment. Predicting consumption before projects go live remains considerably more difficult. That’s becoming increasingly important as AI moves beyond experimentation and into day-to-day operations.
Oliver Nowak, AI advisory consultant at ServiceNow Elite Partner Crossfuze, said customers are only beginning to understand what consumption-based AI means in practice.
“Consumption is now real money rather than a roadmap conversation,” he said.
He believes many organizations are still focusing on exciting AI use cases without fully understanding the long-term consumption profile behind them.
“The biggest blind spot I see is customers get pulled in by the generative and agentic AI capability and start scoping use cases without ever stepping back to ask what the consumption profile actually looks like.”
Rather than discouraging AI adoption, he argued that organizations should build consumption modelling into project planning from the outset, enabling them to understand likely costs before deployment rather than reacting with shock and fear once budgets begin to increase.
Governance Enters a New Phase
For the past couple of years, enterprise AI governance has focused on familiar concerns such as security, compliance, data protection and model behaviour. And while those priorities remain in place, financial governance has now entered the chat.
For Spearing, organizations moving AI into production typically have two major concerns.
“The first one is, am I exposing the company to a hell of a lot of risk… and the second one is, is my CFO going to jump all over me because suddenly my bills got up 10x?”
It’s a shift that reflects broader market trends. Julie L. Mohr, principal analyst at Forrester, said governance remains the primary concern among enterprise customers.
“Their primary concern is really overall governance,” Mohr recently told NowBen. “Clients need transparency on their demand and spend in addition to understanding how it is driving value and outcomes.”
As organizations deploy increasingly autonomous AI capabilities, she added, “observability of the agentic AI is a high priority.”
Those priorities closely mirror the direction ServiceNow is taking with AI Control Tower. Spearing said the platform is intended to provide organizations with greater visibility into AI usage while also giving them stronger operational controls.
“In August you get effectively the kill switch with AI,” he said. “We can say that agent’s gone rogue, shut it off… that language model has gone rogue, close it down.”
Spearing did reveal, however, that predictive consumption capabilities are part of the platform’s future direction, so watch this space.
A Changing Role for Partners
The shift towards operational AI is also changing expectations of ServiceNow partners.
Historically, partners have focused on implementation, integration, and platform expansion. Increasingly, customers are looking for advice on governance, forecasting, consumption modelling and operational best practice.
Nowak believes that organizations should be selective rather than indiscriminate in their use of AI.
“Customers should be using AI where it makes a tangible, material difference, not just for the sake of using it,” he said.
He also warned against organizations focusing solely on limiting costs.
“The customers I worry about are not the cautious ones; they’re the over-enthusiastic ones… The risk in the other direction is finance teams becoming too conservative… and starving AI of the runway to prove itself.”
Spearing also believes that evolution creates a new opportunity for partners.
“Partners have got to think about their own reinvention as well,” he said. “If you are proactive with your customers… and go, ‘Look, we know the AI is going to affect the cost model, let’s have a new conversation,’ you’ll be in a very different place in two years’ time.”
Rather than just advising on licensing, partners are increasingly helping organizations understand where AI creates measurable value, how it should be governed, and how operational costs can be monitored as adoption grows.
Final Thoughts
Ultimately, the new ServiceNow pricing model signifies more than a commercial overhaul.
By embedding AI more deeply into the platform and introducing consumption-based pricing, the company has accelerated a broader shift already taking place across enterprise IT. Organizations are no longer deciding whether to adopt AI. They’re learning how to forecast it, govern it, measure its value and operate it at scale.
ServiceNow won’t be the last enterprise software vendor to make this transition. As AI becomes embedded across business applications, the challenge of governing consumption rather than simply purchasing licences is likely to become a defining issue for enterprise IT over the next few years.